This invention relates to epitaxial growth processes and, more particularly, to epitaxial growth on compound semiconductor substrates containing a relatively volatile element. Illustratively, these substrates are Group III-V compounds such as InP or GaAs which are commonly used in the fabrication of GaAs/AlGaAs or InGaAsP/InP semiconductor devices; e.g., lasers and LEDs, photodiodes and FETs. The lasers and LEDs typically have a double heterostructure (DH) geometry which includes a narrow bandgap active layer sandwiched between a pair of wider bandgap cladding layers; e.g., AlGaAs-GaAs-AlGaAs or InP-InGaAsP-InP.
Due to the low absorption coefficient of optical fibers in the 1.3-1.6 .mu.m wavelength range, InGaAsP/InP DH lasers and LEDs have become extremely promising light sources for light-wave communications. To date, liquid phase epitaxy (LPE) has been the predominant method used to fabricate InGaAsP multilayer structures on InP substrates, but even this growth technique is not perfect, being subject to myriad defects which render devices irresproducible. Among the possible causes of irreproducibility in LPE growth, thermal decomposition of InP substrates is of prime importance. A similar problem exists in vapor phase epitaxy (VPE) and molecular beam epitaxy growth on InP substrates.
In LPE, for example, the loss of phosphorus from the substrates at the high temperatures present during heating cycles results in an indium rich region which in turn creates a perturbed epitaxial layer. Moreover, the etched mesas of buried heterostructures lasers may be damaged by thermal degradation prior to regrowth.
Another problem related to the LPE growth of InGaAsP alloys is the stability of the solid solution epitaxial layer. It has been suggested that quaternary InGaAsP solid solutions may be metastable, resulting in spinoidal decomposition, even at low temperature. In such a case, the growth of other layers after the quaternary active layer, as well as any thermal treatment, could lead to an irreproducible precipitation of the spinoidal solid solutions. This problem may be even further complicated by the presence of a surface strain field due to thermal decomposition, prior to growth of the quaternary active layer. Controlling the thermal degradation of the InP substrates before growth becomes necessary to eliminate the irreproducibility induced by the interface strain field.
One method proposed in the prior LPE art to remove the damaged substrate surface is an in situ etching step, or melt-back technique, described by V. Wrick et al in Electronics Letters, Vol. 12, p. 394 (1976). This method, however, leads to nonplanar surface morphology and cannot be used when micron size features, such as etched mesas or channels, are present on the wafer.
Other techniques have been proposed and used with varying degrees of success to prevent thermal decomposition during LPE. First, the use of an InP cover wafer in close proximity to the InP substrate has been shown to significantly reduce the density of etch pits resulting from thermal etching and degradation. See, A. Doi et al, Applied Physics Letters, Vol. 34, p. 393 (1979). However, this method suffers from an inherent limitation because the phosphorus pressure in equilibrium with the InP substrate is merely contained, not increased.
Alternative methods were recently proposed in LPE for providing excess phosphorus in the vicinity of the InP substrate by using elemental phosphorus powder, M. A. DiGiuseppe et al, Journal of Crystal Growth, Vol. 58, p. 279 (1982), or phosphine (PH.sub.3) gas, A. R. Clawson et al, Journal of Crystal Growth, Vol. 46, p. 300 (1979). Both methods, however, introduce phosphorus in the entire growth system rather than near the InP substrate only. Moreover, the elemental phosphorus method gave results comparable to the cover waver method as measured by the ratios of integrated photoluminescence intensity in the case of S-doped InP, and very poor protection in the case of both Sn-doped InP and undoped InP.
In the case of the phosphine method, the results showed no change in either physical appearance, electrical characteristics, or photoluminescence. Nevertheless, the decomposition of PH.sub.3 into P.sub.2 and P.sub.4 is not complete and safety precautions associated with the use of toxic gases must be observed.
All of these methods introduce phosphorus into the entire growth system and may modify the melt compositions. Consequently, the development of a method providing excess phosphorus localized to the immediate vicinity of the InP substrate is desirable. Recently, G. A. Antypas reported in Applied Physics Letters, Vol. 37, p. 64 (1980) the use of an open graphite basket containing a solution of Sn-In-P for eliminating visible decomposition of InP substrates before and after LPE growth. However, no results of photoluminescence or electrical measurements were provided to substantiate the photomicrographic results.